JP7361219B2 - Defect inspection device, defect inspection method and program, printing device, printed matter manufacturing method - Google Patents

Description

The present invention relates to a defect inspection device, a defect inspection method and program, a printing device, and a method for manufacturing printed matter, and in particular, detects defects in the inspection target by comparing captured image data of the inspection target with reference data. Regarding detection technology.

2. Description of the Related Art Defect inspection apparatuses are widely used that inspect defects in a target by comparing image data obtained by imaging the target using a scanner and reference data serving as a reference. An application of this inspection method is a printed matter defect inspection device for inspecting printed matter printed by a printing device for printing defects such as streaks and ink chips (see Patent Document 1).

In conventional printed matter defect inspection, a widely used method is to obtain reference data in advance, which is imaged data of a reference printed matter printed under the same conditions, and compare it with the imaged data of the printed matter being printed. It is being done. However, this method has a problem in that it is not possible to inspect for defects in printed matter such as variable printing where the pattern changes for each page, or to inspect for defects in the first printed matter after changing the printing conditions.

For example, since the brightness value of the entire data differs depending on the paper, the standard data is obtained from a print using gloss paper, and after inspecting the print on gloss paper, the print on high-quality paper printed with the same pattern is inspected. In this case, there is a large difference between the reference data of the printed material on gloss paper and the imaged data of the printed material on high-quality paper, and if the comparison is made directly, it is necessary to reduce the inspection performance. In addition, in order to perform the inspection without degrading the inspection performance, it is necessary to acquire the reference data again from the printed matter using high-quality paper.

To solve this problem, it is conceivable to perform inspection using print data based on the original image of the printed material as reference data.

Japanese Patent Application Publication No. 07-186375

However, when performing an inspection using print data as reference data, there is a problem in that it is difficult to align the reference data and imaged data.

Conventional alignment methods generally involve matching features within an image, but there is a problem in that the processing becomes slow when applied to a large-sized image such as a captured image of a printed matter. In particular, there is a problem in that the more accurate the alignment is, the more the processing time becomes a trade-off, and the processing becomes slower.

The present invention has been made in view of the above circumstances, and is a defect inspection method that inspects defects in printed matter by aligning imaging data of printed matter based on reference data and reference data with high accuracy and high speed. The purpose of the present invention is to provide a device, a defect inspection method and program, a printing device, and a method for manufacturing printed matter.

One aspect of the defect inspection device for achieving the above object includes an inkjet head in which a plurality of nozzles are arranged in the nozzle direction, and a movement in which the inkjet head and the printing medium are relatively moved in a relative movement direction that intersects with the nozzle direction. A defect inspection device for inspecting defects in printed matter printed based on reference data by a single-pass printing device, comprising: a memory for storing instructions to be executed by a processor; and a processor that executes an instruction to acquire the imaging data and reference data based on the captured image of the printed matter, and determine the positions of the plurality of nozzles and the pixel positions in the nozzle direction of the captured data. Acquire nozzle mapping information that is a correspondence relationship between This is a defect inspection device that performs defect inspection processing to calculate defect information by inputting the information and reference data.

According to this aspect, it is possible to inspect the printed material for defects by aligning the image data of the printed material and the reference data with high accuracy and high speed based on the reference data.

The processor acquires print timing information that is a correspondence relationship between the print timing of the reference data and the pixel position in the relative movement direction of the imaged data, and uses the print timing information to determine the position of the imaged data and the reference data in the relative movement direction. It is preferable to perform a relative movement direction alignment process, and after performing a nozzle direction alignment process and a relative movement direction alignment process, perform a defect inspection process to calculate defect information using the imaging data and reference data as input. Thereby, the positions of the imaging data and the reference data in the relative movement direction can be matched with high precision and at high speed.

It is preferable that the relative movement direction alignment process includes a start timing calculation process of calculating the printing start timing of the reference data among the printing timing information. Thereby, the positions of the imaging data and the reference data in the relative movement direction can be appropriately aligned.

It is preferable that the start timing calculation process analyzes a captured chart image in which a correction chart having a reference line is captured, and uses the position of the reference line. Thereby, the printing start timing of the reference data can be appropriately calculated.

It is preferable that the print timing information is information based on an encoder value of the moving mechanism. Thereby, print timing information can be appropriately acquired.

Preferably, the encoder value is embedded in the captured image. Thereby, print timing information can be appropriately acquired.

Preferably, the nozzle mapping information includes at least one of a plurality of pieces of information depending on the thickness of the print medium and information corrected by a correction process depending on the thickness. Thereby, nozzle mapping information can be appropriately acquired.

It is preferable that the processor performs defect inspection processing using a deep learning model inputted with imaging data and reference data. This embodiment is suitable for defect inspection processing using a deep learning model.

One aspect of the printing device for achieving the above object includes an inkjet head in which a plurality of nozzles are arranged in the nozzle direction, and a movement mechanism that relatively moves the inkjet head and the print medium in a relative movement direction that intersects with the nozzle direction. The present invention is a single-pass printing device comprising: a scanner that images a printed matter to generate a captured image; and the defect inspection device described above.

According to this aspect, it is possible to inspect the printed matter for defects by aligning the imaging data of the printed matter printed by the inkjet head and the reference data with high accuracy and high speed based on the reference data.

One aspect of the defect inspection method for achieving the above object is to move an inkjet head in which a plurality of nozzles are arranged in the nozzle direction, and to relatively move the inkjet head and the printing medium in a relative movement direction that intersects with the nozzle direction. A defect inspection method for inspecting defects in printed matter printed based on reference data by a single-pass printing device comprising: a nozzle mapping information acquisition step that acquires nozzle mapping information that is a correspondence between the positions of a plurality of nozzles and pixel positions in the nozzle direction of the imaged data, and a data acquisition step that acquires the imaged data using the nozzle mapping information A nozzle direction alignment process that performs a nozzle direction alignment process to match the nozzle direction position with reference data, and a defect inspection that calculates defect information by inputting imaging data and reference data after performing the nozzle direction alignment process. A defect inspection method includes a defect inspection step of performing processing.

According to this aspect, it is possible to inspect the printed material for defects by aligning the image data of the printed material and the reference data with high accuracy and high speed based on the reference data.

One aspect of the method for manufacturing printed matter to achieve the above object is to relatively move an inkjet head in which a plurality of nozzles are arranged in the nozzle direction, and the inkjet head and the print medium in a relative movement direction that intersects with the nozzle direction. a printing process of printing a printed matter based on reference data by a single-pass printing device comprising a moving mechanism; a defect inspection method described above; and a quality determination process of determining the quality of the printed matter based on defect information. A method of manufacturing a printed matter, comprising:

According to this aspect, it is possible to appropriately determine the quality of printed matter.

One aspect of the program for achieving the above object is a program for causing a computer to execute the defect inspection method described above. This embodiment may also include a computer-readable non-transitory storage medium on which this program is recorded.

According to this aspect, it is possible to inspect the printed material for defects by aligning the image data of the printed material and the reference data with high accuracy and high speed based on the reference data.

According to the present invention, it is possible to inspect the printed matter for defects by aligning the image data of the printed matter with the reference data with high accuracy and high speed based on the reference data, so that inspection performance due to positional deviation can be inspected. The decline can be prevented.

FIG. 1 is a block diagram showing the configuration of a defect inspection device. FIG. 2 is an overall configuration diagram of the inkjet printing apparatus. FIG. 3 is a plan view showing the nozzle surface of the inkjet head. FIG. 4 is a plan view showing the reading surface of the scanner. FIG. 5 is a block diagram showing the configuration of the control system of the inkjet printing apparatus. FIG. 6 is a diagram showing an example of imaging data and reference data. FIG. 7 is a diagram showing an example of imaging data and reference data. FIG. 8 is a diagram showing an example of imaging data and reference data. FIG. 9 is a flowchart showing the processing of the printed matter manufacturing method.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

<Configuration of defect inspection device>
FIG. 1 is a block diagram showing the configuration of a defect inspection apparatus 10. As shown in FIG. The defect inspection device 10 is a device that inspects defects in printed matter printed based on reference data by a single-pass printing device (for example, the inkjet printing device 100 shown in FIG. 2). The defect inspection apparatus 10 inspects the printed matter for defects by comparing image data based on a captured image of the printed matter using an imaging device (for example, the scanner 132 shown in FIG. 2) with reference data. As shown in FIG. 1, the defect inspection apparatus 10 includes a processor 12 and a memory 14.

Processor 12 executes instructions stored in memory 14. Memory 14 stores instructions for execution by processor 12.

The hardware structure of the processor 12 includes the following types of processors. Various processors include a CPU (Central Processing Unit), which is a general-purpose processor that executes software (programs) and functions as various processing units, a GPU (Graphics Processing Unit), which is a processor specialized in image processing, Programmable Logic Devices (PLDs), which are processors whose circuit configuration can be changed after manufacturing, such as FPGAs (Field Programmable Gate Arrays), and ASICs (Application Specific Integrated Circuits), which are dedicated to executing specific processes. This includes a dedicated electrical circuit that is a processor with a designed circuit configuration.

The processor 12 may be configured with one of these various types of processors, or may be configured with two or more processors of the same or different types (for example, multiple FPGAs, a combination of a CPU and an FPGA, or a combination of a CPU and a GPU). combination).

Furthermore, the hardware structure of these various processors is, more specifically, an electric circuit (circuitry) that is a combination of circuit elements such as semiconductor elements.

The processor 12 includes an alignment processing section 16. Imaging data and reference data are input to the alignment processing section 16 . The alignment processing unit 16 performs an alignment process between the imaging data and the reference data, and outputs aligned image data consisting of the aligned imaging data and the reference data. The process of aligning the image data and the reference data refers to obtaining the correspondence between each pixel of the image data and each pixel of the reference data.

The alignment processing section 16 includes a nozzle direction alignment processing section 18 , a nozzle mapping information holding section 20 , a transport direction alignment processing section 22 , and a print timing information holding section 24 .

The nozzle direction positioning processing unit 18 performs nozzle direction positioning processing to match the positions of the imaging data and the reference data in the nozzle direction. That is, the nozzle direction alignment processing unit 18 acquires the correspondence relationship between each pixel of the image data and each pixel of the reference data in the nozzle direction.

The nozzle mapping information holding unit 20 holds nozzle mapping information necessary for nozzle direction alignment processing. The nozzle mapping information holding unit 20 may be provided in the memory 14. Details of the nozzle mapping information will be described later.

The conveyance direction alignment processing unit 22 performs a conveyance direction alignment process (an example of a relative movement direction alignment process) to align the captured data and the reference data in the conveyance direction (an example of a relative movement direction intersecting the nozzle direction). conduct. That is, the conveyance direction alignment processing unit 22 acquires the correspondence relationship between each pixel of the imaged data and each pixel of the reference data in the conveyance direction.

The print timing information holding unit 24 holds print timing information necessary for the conveyance direction alignment process. The print timing information holding section 24 may be provided in the memory 14. Details of the print timing information will be described later.

The processor 12 also includes a defect inspection processing section 26 . The image data after alignment is input to the defect inspection processing section 26 . The defect inspection processing unit 26 performs defect inspection processing to calculate defect information by inputting the image data after alignment, and outputs the calculated defect information.

The defect inspection processing unit 26 includes a deep learning model 28. The deep learning model 28 is a trained model that receives image data and reference data as input and outputs defect information on printed matter. The deep learning model 28 has a plurality of layer structures and holds a plurality of weight parameters. The deep learning model 28 can change from an unlearned model to a learned model by updating the weight parameters from initial values to optimal values. The defect information includes at least one of the presence or absence of a defect, the position of the defect, and the recognition strength value of the defect.

<Printing device configuration>
Next, a single-pass printing apparatus to which the defect inspection apparatus 10 is applied will be described. FIG. 2 is an overall configuration diagram of the inkjet printing apparatus 100. In FIG. 2, the X direction, the Y direction, and the Z direction are directions perpendicular to each other, the X direction and the Y direction are horizontal directions, and the Z direction is a vertical direction. The inkjet printing device 100 prints a color image on a sheet of paper P, which is a print medium, by ejecting ink in four colors: cyan (C), magenta (M), yellow (Y), and black (K). It's a printing machine.

As the paper P, a general-purpose printing paper is used. General-purpose printing paper refers to paper mainly made of cellulose, such as coated paper used for general offset printing, etc., rather than so-called inkjet paper. Furthermore, water-based ink is used as the ink. Water-based ink refers to ink in which coloring materials such as dyes and pigments are dissolved or dispersed in water and a water-soluble solvent.

As shown in FIG. 2, the inkjet printing apparatus 100 includes a conveying section 110, a printing section 120, an imaging section 130, a drying section 140, a sorting section 150, and a paper discharge section 160.

[Transportation section]
The transport unit 110 is a transport mechanism that transports paper P fed from a paper feed unit (not shown) in the transport direction (Y direction). The transport unit 110 corresponds to a moving mechanism that relatively moves the printing unit 120 and the paper P in a relative movement direction. The conveyance unit 110 includes an upstream pulley 112, a rotary encoder 113, a downstream pulley 114, and a conveyance belt 116.

The upstream pulley 112 has a rotating shaft (not shown) extending in the horizontal direction, and the rotating shaft is rotatably supported. A rotary encoder 113 is arranged on the upstream pulley 112. The rotary encoder 113 outputs an encoder value according to the rotation of the upstream pulley 112.

The downstream pulley 114 has a rotating shaft (not shown) that is parallel to the rotating shaft of the upstream pulley 112, and the rotating shaft is rotatably supported.

The conveyor belt 116 is an endless belt made of stainless steel. The conveyor belt 116 spans the upstream pulley 112 and the downstream pulley 114. The conveyance unit 110 can maintain good flatness of the paper P by using a conveyor belt 116 made of stainless steel.

The downstream pulley 114 has a motor (not shown) as a driving means. When the motor is driven, the downstream pulley 114 rotates counterclockwise in FIG. 2 . The upstream pulley 112 rotates counterclockwise in FIG. 2 following the rotation of the downstream pulley 114. Due to the rotation of the upstream pulley 112 and the downstream pulley 114, the conveyor belt 116 travels between the upstream pulley 112 and the downstream pulley 114 along the travel path.

On the conveying surface of the conveying belt 116, paper P supplied from a paper feeding section (not shown) is placed. The conveyance unit 110 conveys the paper P placed on the conveyance belt 116 along a conveyance path from the upstream pulley 112 to the downstream pulley 114 and delivers it to the paper discharge unit 160. At a position on this conveyance path facing the printing section 120, the imaging section 130, the drying section 140, and the sorting section 150, the paper P is conveyed with its printing surface held horizontally.

The conveyor belt 116 is provided with a plurality of suction holes (not shown), and by suctioning the suction holes of the conveyor belt 116 using a pump (not shown), the paper P placed on the conveyance surface of the conveyor belt 116 is suctioned and held on the conveyance surface. You may.

[Printing Department]
The printing unit 120 prints an image on paper P based on the reference data. The printing unit 120 includes inkjet heads 122C, 122M, 122Y, and 122K. The inkjet head 122C ejects cyan ink droplets using an inkjet method. Similarly, the inkjet heads 122M, 122Y, and 122K each eject magenta, yellow, and black ink droplets using an inkjet method.

The inkjet heads 122C, 122M, 122Y, and 122K are arranged at regular intervals along the conveyance path of the paper P by the conveyance belt 116. The inkjet heads 122C, 122M, 122Y, and 122K are line heads each having a length corresponding to the paper width. The inkjet heads 122C, 122M, 122Y, and 122K are arranged such that the nozzle surface 124 (see FIG. 3) faces the conveyor belt 116.

The inkjet heads 122C, 122M, 122Y, and 122K eject ink droplets from a plurality of nozzles 126 (see FIG. 3) arranged on the nozzle surface 124 toward the paper P conveyed by the conveyance belt 116. An image is printed on the printing surface of paper P with a predetermined mesh type. The timing at which each of the inkjet heads 122C, 122M, 122Y, and 122K ejects ink droplets is synchronized with the encoder value obtained from the rotary encoder 113 disposed on the upstream pulley 112.

In this way, the printing unit 120 generates printed matter by scanning the paper P conveyed in the Y direction by the conveyance belt 116 once, using a so-called single-pass method.

FIG. 3 is a plan view showing the nozzle surface 124 of the inkjet head 122C. As shown in FIG. 3, a plurality of nozzles 126 are arranged on the nozzle surface 124 in the nozzle direction (X direction). Although FIG. 3 shows an example in which the plurality of nozzles 126 are lined up in one row in the X direction for the sake of simplification, the plurality of nozzles 126 may be two-dimensionally arranged on the nozzle surface 124. The plurality of nozzles 126 arranged two-dimensionally are substantially one nozzle row (projected nozzle row) orthogonally projected on a straight line along a direction perpendicular to the direction of relative movement between the inkjet head 122C and the paper P. Configure columns. In this embodiment, the direction perpendicular to the direction of relative movement between the inkjet head 122C and the paper P can be defined as the nozzle direction.

The configurations of the inkjet heads 122M, 122Y, and 122K are also similar to the inkjet head 122C.

[Imaging section]
Returning to the description of FIG. 2, the imaging unit 130 acquires an image of the printing surface of the paper P. The imaging unit 130 is arranged downstream of the printing unit 120 with respect to the conveyance direction of the paper P. The imaging unit 130 includes a scanner 132.

The scanner 132 is arranged so that a reading surface 134 (see FIG. 4) faces the conveyor belt 116. The scanner 132 is a line sensor in which a plurality of light receiving elements 136 (see FIG. 4) are arranged side by side in the X direction on a reading surface 134. As the line sensor, for example, a CCD (Charge Coupled Device) sensor or a CMOS (Complementary Metal Oxide Semiconductor) sensor is used. The scanner 132 optically reads images printed on paper P using the inkjet heads 122C, 122M, 122Y, and 122K with a plurality of light receiving elements 136, and generates image data based on the captured images.

FIG. 4 is a plan view showing the reading surface 134 of the scanner 132. As shown in FIG. 4, a plurality of light receiving elements 136 are arranged on the reading surface 134 in the nozzle direction (X direction). Here, the reading resolution of the scanner 132 is lower than the printing resolution of each of the inkjet heads 122M, 122Y, and 122K.

The imaging unit 130 may include a light source that irradiates the image printed on the paper P with illumination light.

[Drying section]
The drying section 140 dries the ink on the paper P. The drying section 140 is arranged downstream of the imaging section 130 with respect to the conveyance direction of the paper P.

The drying section 140 includes a heater 142. As the heater 142, for example, at least one of a halogen heater and an infrared heater is used. The heater 142 heats the printing surface of the paper P to dry the ink on the paper P. The drying section 140 may include blowing means such as a fan or a blower.

[Sorting Department]
The sorting unit 150 determines the quality of the printed matter according to defect information regarding the paper P conveyed by the conveyance belt 116, and sorts the printed matter. The sorting section 150 is arranged downstream of the drying section 140 with respect to the conveyance direction of the paper P. The sorting section 150 includes a stamper 152.

The stamper 152 performs a stamping process in which ink is applied to the leading edge of the paper P that is determined to be a defective print, depending on the quality determination regarding the paper P transported by the transport belt 116 .

[Paper output section]
The paper discharge unit 160 collects paper P (printed material) on which an image has been printed and dried. The paper discharge section 160 is disposed downstream of the sorting section 150 with respect to the conveyance direction of the paper P, and at the end point of the conveyance path of the conveyance section 110. The paper discharge section 160 includes a paper discharge tray 162.

The paper discharge tray 162 stacks and collects the sheets P transported by the transport belt 116. The paper ejection tray 162 is provided with a front paper rest, a rear paper rest, and a horizontal paper rest (not shown), and stacks the sheets P in an orderly manner.

Further, the paper ejection tray 162 is provided so as to be movable up and down by a lifting device (not shown). The driving of the elevating device is controlled in conjunction with the increase/decrease in the number of sheets P stacked on the paper ejection tray 162. As a result, the highest sheet P among the sheets P stacked on the sheet discharge table 162 always has a constant height.

The inkjet printing apparatus 100 transports the paper P and moves the inkjet heads 122C, 122M, 122Y, and 122K relative to the paper P. The inkjet heads 122C, 122M, 122Y, and 122K and the paper P may be moved relative to each other.

[Control system of inkjet printing device]
FIG. 5 is a block diagram showing the configuration of the control system of the inkjet printing apparatus 100. The inkjet printing device 100 includes a defect inspection device 10, a user interface 170, a storage section 172, an overall control section 174, a conveyance control section 176, a printing control section 178, an imaging control section 180, and a drying control section 182. , a sorting control section 184 , and a paper discharge control section 186 .

The user interface 170 includes an input section (not shown) for the user to operate the inkjet printing apparatus 100, and a display section (not shown) for presenting information to the user. The input unit is, for example, an operation panel that accepts input from the user. The display unit is, for example, a display that displays image data and various information. By using the user interface 170, the user can cause the inkjet printing apparatus 100 to print a desired image.

The storage unit 172 stores a program for controlling the inkjet printing apparatus 100 and information necessary for executing the program. The storage unit 172 is configured by a non-temporary storage medium such as a hard disk (not shown) or various semiconductor memories. The storage unit 172 may store reference data. The defect inspection apparatus 10 may acquire the reference data from the storage unit 172.

The overall control unit 174 performs various processes according to programs stored in the storage unit 172, and controls overall operations of the inkjet printing apparatus 100.

The conveyance control unit 176 causes the conveyance unit 110 to convey the paper P in the conveyance direction by controlling a motor (not shown) of the conveyance unit 110 . As a result, the paper P supplied from a paper feed section (not shown) passes through a position facing the printing section 120, the imaging section 130, the drying section 140, and the sorting section 150, and finally the paper discharge section. The paper is ejected at 160. Further, the conveyance control unit 176 acquires an encoder value from the rotary encoder 113.

The print control unit 178 controls ink ejection by the inkjet heads 122C, 122M, 122Y, and 122K based on reference data based on the printing source image. The print control unit 178 causes cyan, magenta, yellow, and black ink droplets to be applied to the paper P by the inkjet heads 122C, 122M, 122Y, and 122K, respectively, in synchronization with the encoder value acquired via the conveyance control unit 176. Dispense toward the target. As a result, a color image is printed on the printing surface of the paper P, and the paper P becomes a "printed material".

Note that the print control unit 178 may output information on the location of the nozzle 126 having a discharge failure to the overall control unit 174. The defect inspection apparatus 10 acquires information about the location of the nozzle 126 having an ejection failure from the print control unit 178.

Furthermore, the print control unit 178 may have a compensation function that compensates for printing by the nozzle 126 with ejection failure by correcting the printing source image. As an example, there is a compensation function that compensates for a nozzle 126 having an ejection failure by increasing the volume of ink droplets of a plurality of adjacent nozzles 126. The print control unit 178 outputs information on the portions of printed matter compensated by the compensation function to the overall control unit 174. The defect inspection device 10 acquires information about the portions of the printed matter compensated for by the compensation function from the print control unit 178.

The imaging control unit 180 controls imaging by the scanner 132, thereby causing the imaging unit 130 to read the image of the paper P (printed material). The imaging control unit 180 causes the scanner 132 to read the image printed on the paper P in synchronization with the encoder value acquired via the conveyance control unit 176. The defect inspection apparatus 10 acquires image data based on the image read by the scanner 132.

The inkjet printing apparatus 100 may cause the print control unit 178 to print a nozzle failure detection pattern, and acquire information about the location of the nozzle 126 with the ejection failure by analyzing the captured image read by the scanner 132.

The drying control unit 182 causes the drying unit 140 to dry the paper P by controlling the heating by the heater 142. The heater 142 heats the paper P when the paper P passes through a position facing the heater 142 .

The sorting control section 184 causes the sorting section 150 to sort the sheets P by controlling stamp processing by the stamper 152. The sorting control unit 184 classifies the printed matter into non-defective printed matter and defective printed matter according to the defect information output from the defect inspection device 10. If the paper P passing through the position facing the stamper 152 is determined to be a defective printed product, the sorting control unit 184 performs stamp processing using the stamper 152 .

The paper discharge control unit 186 controls the stacking of paper P on the paper discharge table 162. The sheets P are discharged onto the paper discharge tray 162 and stacked. Ink is attached to the leading edge of the paper P of the defective printed product. Therefore, the user can identify defective printed products from among the sheets P stacked on the paper output table 162.

<Nozzle mapping information>
The nozzle mapping information held by the nozzle mapping information holding unit 20 indicates the correspondence between the position of a pixel of imaged data (pixel position) in the nozzle direction and the position of each nozzle 126 of the inkjet heads 122C, 122M, 122Y, and 122K. This is the information shown. That is, the nozzle mapping information indicates which pixel of the image data based on the captured image read by the scanner 132 is imaged with a dot ejected from which nozzle 126 .

Here, for the sake of explanation, nozzle numbers 1, 2, 3, etc. are assigned to the plurality of nozzles 126 of each of the inkjet heads 122C, 122M, 122Y, and 122K in order from one nozzle direction to the other. .

For example, by continuously ejecting ink from the plurality of nozzles 126 of the inkjet head 122C that are sufficiently spaced apart in the captured image, the paper P can be printed with cyan line segments extending in the transport direction. is printed. By acquiring image data obtained by reading this line segment with the scanner 132 and detecting which light receiving element 136 read the printed line segment, the nozzle number of the nozzle 126 that ejected the ink can be determined for the inkjet head 122C. The correspondence relationship between each pixel of the image data can be obtained.

In this process, ink is ejected from one nozzle 126 for every 100 nozzles 126 of the inkjet head 122C, and a plurality of line segments extending in the transport direction at equal intervals in the nozzle direction are printed. By reading the plurality of line segments with the scanner 132, it becomes possible to estimate the position of the light receiving element 136 in the image data at which the line segment by each nozzle 126 of the inkjet head 122C is printed. The same applies to inkjet heads 122M, 122Y, and 122K.

The nozzle mapping information is held, for example, as a table showing all the pixels of the image data to which the line segments formed by each nozzle 126 correspond. The nozzle mapping information may hold pixel positions of imaging data for nozzles 126 at regular intervals. The nozzle mapping information held in this manner may be converted into a linear form and used.

Moreover, the nozzle mapping information may hold only information about which pixel position of the image data corresponds to a line segment formed by the nozzles 126 at both ends in the nozzle direction. In the nozzle mapping information held in this manner, the pixel position of the imaged data for each nozzle 126 may be interpolated on the assumption that the nozzles 126 between both ends are all equally spaced.

In the single-pass type inkjet printing apparatus 100, the nozzle 126 that outputs each pixel in the nozzle direction in the reference data is always fixed. Therefore, it is possible to determine from which position of the plurality of nozzles 126 the reference data is output. On the other hand, which pixel position in the imaging data corresponds to the line segment formed by each nozzle 126 can be determined from the nozzle mapping information. Therefore, it is possible to determine which pixel position of the image data corresponds to a pixel in the nozzle direction of the reference data.

FIG. 6 is a diagram showing an example of the imaging data _DS and the reference data _DR . As shown in FIG. 6, in this example, regarding the reference data D _R , the left end in the nozzle direction is printed by the nozzle 126 with nozzle number 2, and the right end in the nozzle direction is printed by the nozzle 126 with nozzle number 1000. .

Here, according to the nozzle mapping information, the 5th pixel of the image data corresponds to the area printed by the nozzles 126 with nozzle numbers 2 and 3, and the 100 pixel of the image data corresponds to the area printed by the nozzles 126 with nozzle numbers 999 and 1000. It is known that it corresponds to the area printed in . Between these, it is sufficient to interpolate to make the pixels and nozzle numbers correspond.

In this way, by using the nozzle mapping information, it is possible to align the imaging data D _S and the reference data D _R in the nozzle direction.

As described above, in the inkjet printing apparatus 100, a nozzle failure detection pattern is printed and the scanner There is a method of performing nozzle inspection by analyzing the captured image read in step 132. At this time, in order to specify the nozzle 126, it is necessary to save a correspondence table showing which nozzle 126 prints at which pixel position in the captured image. Nozzle mapping information based on this correspondence table may be used.

Note that the nozzle mapping information is preferably changed depending on the thickness of the paper P. When a fixed scanner 132 images a printed matter, a difference in the thickness of the paper P causes a slight deviation from the focal length of the light receiving element 136, causing expansion/contraction within the captured image, causing a shift in nozzle mapping information. Therefore, by holding multiple pieces of nozzle mapping information according to the thickness of the paper P, or by performing a correction process that corrects the nozzle mapping information according to the thickness of the paper P, more accurate positioning can be achieved. It becomes possible to do so.

<Print timing information>
The print timing information held by the print timing information holding unit 24 indicates the correspondence between the pixel position of the image data in the transport direction and the print (print) timing of the reference data in each of the inkjet heads 122C, 122M, 122Y, and 122K. It is information. That is, the print timing information indicates which pixel and at which timing a dot ejected from the nozzle 126 is imaged in the image data based on the image read by the scanner 132.

The inkjet printing apparatus 100 may print a line segment extending in the nozzle direction based on the encoder value of the rotary encoder 113. In this case, print timing information can be obtained by determining the correspondence between the print timing of each line segment of the reference data and the pixels of each line segment of the imaged data using the encoder value.

FIG. 7 is a diagram showing an example of the imaging data D _S and the reference data D _R. As shown in FIG. 7, in this example, the encoder value when printing the leading edge in the transport direction of the reference data _DR is 2, and the encoder value when printing the trailing edge is 1000.

Here, according to the print timing information, the 5th pixel from the leading edge in the transport direction of the captured image data D _S corresponds to the printed area with the encoder value of 2 in the reference data _DR , and the 100th pixel from the leading end in the transport direction is the standard. It is known that the encoder value of the data _DR is 1000 and corresponds to the printed area. Between these, interpolation may be performed to match the pixels and encoder values.

In this way, by using the print timing information, it is possible to align the image data D _S and the reference data D _R in the transport direction.

The print timing information may use values other than encoder values. For example, in the case of the inkjet printing apparatus 100 that prints at a constant clock from the print start timing, the elapsed time from the print start timing may be used as the print timing information.

Furthermore, the print timing information may be a separate profile from the captured image, or may be retained in a form that is added to the captured image. When a line sensor is used as the scanner 132 as in this embodiment, each line is processed at high speed. ), it is lower cost and lower risk to embed it in the image.

When printing of the reference data is started at the print start timing, it is possible to align the imaging data and the reference data in the transport direction by matching the print start timing. However, the relationship between the reading start position of the scanner 132 and the reference data printing start position differs depending on the mounting position of the scanner 132, and therefore varies from machine to machine due to variations in mounting dimensions at the time of installation. In this case, it is necessary to calculate the printing start position of the reference data or convert it into absolute information.

As a specific example of how to calculate the print start position, the print start timing of the reference data is calculated based on the position of the reference line of the imaging data calculated when analyzing the correction pattern and the print start timing of the reference line. (an example of a start timing calculation process that calculates the printing start timing of reference data) is given as an example. When analyzing a correction chart performed before printing or a correction pattern printed during printing, the reference line for determining the analysis position of the correction pattern can be found using the coordinates of the captured image (chart captured image). The offset value (correction coefficient) can be determined based on the encoder value when the reference line is printed and the encoder value when the reference line position of the imaging data is read. Thereby, it is possible to obtain print timing information in which the print start timings of the reference data and the imaged data are synchronized.

FIG. 8 is a diagram showing an example of the imaging data D _S and the reference data D _R. As shown in FIG. 8, in this example, the test pattern T is printed on the upstream side in the conveyance direction of the paper P in the printing area of the reference data _DR .

The reading start position of the scanner 132 and the start of the transport direction of the test pattern T vary depending on the dimensions of the machine and the like. Therefore, the relationship between the start position of the test pattern T in the transport direction and the start position of the imaged data _DS in the transport direction is calculated from the encoder value at the time of printing.

The imaging data of the test pattern T is analyzed, and it can be determined which position in the imaging data is the start position of the test pattern T from the reference line position of the test pattern T (the beginning of the test pattern T ). This allows the correspondence between the encoder values in the imaged data and the encoder values at the time of printing to be known.

For example, assume that printing of the reference data _DR is started from a position where the encoder value is advanced by 20 relative to the start position of the test pattern T. In this case, by detecting the start position of the test pattern T from the image data, the print timing information determines that the reference data _DR is printed at a position where the encoder value in the image data advances by 20 from the start position of the test pattern T. You can see that this is the start of the area.

The test pattern T may be printed after the reference data _DR is printed. That is, the test pattern T may be placed downstream of the printing area of the reference data _DR .

<Manufacturing method of printed matter>
FIG. 9 is a flowchart showing the processing of the printed matter manufacturing method. The printed matter manufacturing method is stored in the storage unit 172 as a program to be executed by a computer.

In step S1 (an example of a data acquisition step), the alignment processing unit 16 of the processor 12 acquires reference data based on a printing source image of a printed matter to be inspected for defects from the storage unit 172. The alignment processing unit 16 may acquire the reference data from the memory 14. The alignment processing unit 16 may acquire the reference data before starting printing of the paper P.

In step S2 (an example of a printing process), the inkjet printing apparatus 100 prints a printed material based on the reference data acquired in step S1.

In step S3 (an example of a data acquisition step), the alignment processing unit 16 acquires imaging data of a printed matter to be inspected for defects. That is, the imaging control unit 180 causes the scanner 132 to read the image printed on the paper P in synchronization with the encoder value acquired via the transport control unit 176. The alignment processing unit 16 acquires the captured image read by the scanner 132 as image data.

In step S4 (an example of a nozzle mapping information acquisition step), the nozzle direction alignment processing section 18 of the alignment processing section 16 acquires nozzle mapping information from the nozzle mapping information holding section 20. The nozzle direction alignment processing section 18 may acquire the nozzle mapping information from the memory 14 or the storage section 172. The defect inspection device 10 prints a nozzle failure detection pattern on the paper P in advance, acquires nozzle mapping information by analyzing the nozzle failure detection pattern, and causes the nozzle mapping information holding unit 20 to hold the acquired nozzle mapping information. I'll keep it.

In step S5 (an example of a nozzle direction alignment process), the nozzle direction alignment processing unit 18 uses the nozzle mapping information acquired in step S4 to match the reference data acquired in step S1 and the imaging data acquired in step S3. Nozzle direction alignment processing is performed to align the positions of the nozzles in the nozzle direction.

In step S<b>6 (an example of a printing timing information acquisition step), the transport direction alignment processing section 22 of the alignment processing section 16 obtains printing timing information from the printing timing information holding section 24 . The conveyance direction alignment processing section 22 may acquire print timing information from the memory 14 or the storage section 172. The defect inspection device 10 prints a chart having a reference line on the paper P in advance, acquires print timing information by analyzing the position of the reference line, and holds the obtained print timing information in the print timing information holding unit 24. I'll let you.

In step S7 (an example of the conveyance direction alignment process), the conveyance direction alignment processing unit 22 uses the print timing information acquired in step S6 to use the reference data and the imaged data aligned in the nozzle direction in step S4. A conveyance direction alignment process is performed to align the conveyance direction position with the conveyance direction.

In step S8 (an example of a defect inspection process), the defect inspection processing unit 26 of the processor 12 inputs the aligned image data consisting of the aligned imaging data and reference data to the deep learning model 28 to detect defects on the printed matter. Performs defect inspection processing to calculate information, and outputs the calculated defect information. The inkjet printing apparatus 100 performs the processing up to step S8 before the paper P printed in step S2 reaches the sorting section 150.

In step S9 (an example of the quality determination step), the sorting unit 150 determines the quality of the printed material based on the defect information output in step S8, and controls the stamper 152 according to the determination.

This completes the process of the printed matter manufacturing method. Note that in the flowchart shown in FIG. 9, step S1 and steps S3 to S8 constitute the defect inspection method according to the present embodiment. The defect inspection method may be stored in the memory 14 as a program to be executed by a computer, or may be executed by the processor 12.

<Defect inspection using deep learning>
When using print data based on the original image of printed matter as reference data, four channels of cyan (C), magenta (M), yellow (Y), and black (K), and red (R) and green (G) are used as reference data. ) and blue (B) channels, such as different data formats, require preprocessing to match their characteristics.

Regarding the difference between CMYK and RGB, a conventional method has been used to match color characteristics by creating a table for matching color profiles between print data and imaging data. However, color profiles vary depending on the printing conditions mentioned above, and since there are a wide variety of conditions such as differences in printing devices and paper types depending on the country, it would be time-consuming to create separate tables for each. There are problems that are time consuming and complicated.

To solve the above problem, we focused on deep learning, which has been widely used in recent years to automate the extraction of parameters and features. Deep learning is a type of machine learning method, and is known to have much higher performance than conventional machine learning methods in the field of image recognition.

In deep learning, a pooling layer is generally introduced in order to deal with variations in the size and position of recognition targets in learning data. The pooling layer is a layer that performs processing that aggregates the pixel values of neighboring pixels into one and compresses the image size. By introducing multiple pooling layers, recognition targets can be identified regardless of their position in the data. It becomes possible to do so.

Addressing positional variations using a pooling layer is effective, for example, when the recognition target exists near the edge of the learning data or when the size of the recognition target is small, but it is effective when two different data are input. There is a problem that it cannot deal with the positional deviation between two pieces of data. For example, if we consider a learning model that inputs print data and image data and determines defects from the image data, if the positions of the print data and image data are misaligned, it will not be possible to determine whether it is a defect or a pattern. .

On the other hand, according to the defect inspection method using the defect inspection device 10, defects in the printed matter can be inspected by aligning the image data of the printed matter and the reference data with high accuracy and high speed based on the reference data. becomes possible. Therefore, printed matter can be appropriately inspected using deep learning.

<Others>
The technical scope of the present invention is not limited to the scope described in the above embodiments. The configurations and the like in each embodiment can be combined as appropriate between the embodiments without departing from the spirit of the present invention.

10... Defect inspection device 12... Processor 14... Memory 16... Processing section 18... Processing section 20... Nozzle mapping information holding section 22... Processing section 24... Print timing information holding section 26... Defect inspection processing section 28... Deep learning model 100... Inkjet printing device 110...Transportation section 112...Upstream pulley 113...Rotary encoder 114...Downstream pulley 116...Transportation belt 120...Printing section 122C...Inkjet head 122K...Inkjet head 122M...Inkjet head 122Y...Inkjet head 124...Nozzle surface 126 ... Nozzle 130 ... Imaging section 132 ... Scanner 134 ... Reading surface 136 ... Light receiving element 140 ... Drying section 142 ... Heater 150 ... Sorting section 152 ... Stamper 160 ... Paper ejecting section 162 ... Paper ejection table 170 ... User interface 172 ... Storage section 174 ...Overall control section 176...Transport control section 178...Print control section 180...Imaging control section 182...Drying control section 184...Sorting control section 186...Paper discharge control section S1 to S7...Each step of defect inspection method

Claims (13)

A single-pass printing device that includes an inkjet head with a plurality of nozzles arranged in the nozzle direction, and a movement mechanism that relatively moves the inkjet head and print medium in a relative movement direction that intersects the nozzle direction, prints reference data. A defect inspection device that inspects defects in printed matter based on
memory that stores instructions for the processor to execute;
a processor that executes instructions stored in memory;
Equipped with
The processor includes:
acquiring imaging data based on a captured image of the printed matter and the reference data;
obtaining nozzle mapping information that is a correspondence relationship between the positions of the plurality of nozzles and the pixel positions in the nozzle direction of the imaging data;
performing nozzle direction positioning processing to match the positions of the imaging data and the reference data in the nozzle direction using the nozzle mapping information;
After performing the nozzle direction alignment process, using the image data and the reference data as input, performing a defect inspection process to calculate defect information.
Defect inspection equipment. The processor includes:
obtaining print timing information that is a correspondence relationship between the print timing of the reference data and the pixel position in the relative movement direction of the imaged data;
performing relative movement direction positioning processing to match the positions of the imaging data and the reference data in the relative movement direction using the print timing information;
After performing the nozzle direction alignment process and the relative movement direction alignment process, using the image data and the reference data as input, performing a defect inspection process to calculate defect information.
The defect inspection device according to claim 1. The relative movement direction alignment process includes a start timing calculation process of calculating a print start timing of the reference data among the print timing information.
The defect inspection device according to claim 2. The start timing calculation process analyzes a captured chart image in which a correction chart having a reference line is captured, and uses the position of the reference line.
The defect inspection device according to claim 3. The printing timing information is information based on an encoder value of the moving mechanism,
The defect inspection device according to any one of claims 2 to 4. the encoder value is embedded in the captured image;
The defect inspection device according to claim 5. The nozzle mapping information includes at least one of a plurality of pieces of information corresponding to the thickness of the print medium and information corrected by a correction process according to the thickness.
A defect inspection device according to any one of claims 1 to 6. The processor performs the defect inspection process using a deep learning model input with the imaging data and the reference data.
A defect inspection device according to any one of claims 1 to 7. A single-pass printing device comprising: an inkjet head with a plurality of nozzles arranged in a nozzle direction; and a movement mechanism that relatively moves the inkjet head and a printing medium in a relative movement direction intersecting the nozzle direction. ,
a scanner that images the printed matter and generates a captured image;
A defect inspection device according to any one of claims 1 to 8,
Printing device with. A single-pass printing device that includes an inkjet head with a plurality of nozzles arranged in the nozzle direction, and a movement mechanism that relatively moves the inkjet head and print medium in a relative movement direction that intersects the nozzle direction, prints reference data. A defect inspection method for inspecting defects in printed matter based on
a data acquisition step of acquiring imaging data based on a captured image of the printed matter and the reference data;
a nozzle mapping information acquisition step of acquiring nozzle mapping information that is a correspondence relationship between the positions of the plurality of nozzles and the pixel positions in the nozzle direction of the imaging data;
a nozzle direction positioning step of performing a nozzle direction positioning process of aligning the positions of the imaging data and the reference data in the nozzle direction using the nozzle mapping information;
a defect inspection step of performing a defect inspection process of calculating defect information using the imaging data and the reference data as input after performing the nozzle direction alignment process;
A defect inspection method comprising: A single-pass printing device that includes an inkjet head with a plurality of nozzles arranged in the nozzle direction, and a movement mechanism that relatively moves the inkjet head and print medium in a relative movement direction that intersects the nozzle direction, prints reference data. a printing process for printing printed matter based on the
The defect inspection method according to claim 10,
a quality determination step of determining the quality of the printed matter based on the defect information;
A method for manufacturing a printed matter comprising: A program for causing a computer to execute the defect inspection method according to claim 10. A non-transitory computer-readable recording medium, on which the program according to claim 12 is recorded.

Images